Design and Analysis of VNF Scaling Mechanisms for 5G-and-Beyond Networks Using Federated Learning

Abstract

This paper deals with Network Slicing-based 5G Networks. A network slice can be defined as a set of network and virtual network function (VNF) resources deployed across multiple independent infrastructure provider domains. Such multi-domain 5G deployments pose challenges since the domains would not share internal resource allocation details. Slice demands and QoS requirements tend to vary dynamically and must be satisfied by scaling the allocated VNF resources. In this paper, we use the federated learning approach in which the training data remains within the respective domains, while the system learns a shared model by aggregating locally-computed updates. Two state-of-the-art deep learning models, namely Long Short-Term Memory (LSTM) and Gated recurrent units (GRU) are used for forecasting. We present a comparison of the performance of the proposed federated system with the centralized system. Further, synthetic data in each domain has been generated using Generative Adversarial Networks (GAN) to improve the forecasting results. A Python-based discrete event simulator model of the proposed auto-scaling system was written. The experiments were used to study the performance of scaling and non-scaling systems across various workloads, and proactive approach’s effectiveness over reactive scaling. The average queue length was reduced by scaling, compared to a non-scaling system, by around 75% to 82% for different traffic type mixes. The results show that the proactive scaling system outperformed the reactive system in terms of the time required to set up new instances by around 70%.

Keywords

• Federated learning
• time series modeling
• VNF scaling
• multi-domain slicing
• 5G
• network slicing